On Tue, May 17, 2011 at 3:07 PM, kenf_nc <ken.fos...@realestate.com> wrote:
> But I can query Cassandra directly for the documents if I wanted/needed to? > > The data is available in the Solandra Keyspace but I wouldn't recommend accessing it directly. Most people who want todo this end up not storing the document fields in Solandra and instead use a unique id to fetch the data from their custom Cassandra CF. > And, when I need to re-index, I could read from Cassandra, index into Solr, > which will write back to Cassandra overwriting the existing document(s)? > > That sounds really strange, but you can use your source data to index back into solandra. but as i mentioned I wouldn't try accessing the data directly from the Solandra keyspace. > Basically the steps would be, index documents into Solr which would write > to > Cassandra. If I need to update a document, I can query it from Solr OR > query > it from Cassandra, make my modification and re-index it back to Solr which > will update Cassandra. If I need to drop my Solr index and completely > recreate it I could read all documents from Cassandra and index them into > the clean Solr instance, which will update (with no change) the documents > in > Cassandra. If I update a document directly in Cassandra without going thru > Solr indexing, the change would show up on a Solr query of that document, > but the search indexes would not reflect any change. Is all that correct? > > Since cassandra scales writes very well I would suggest you index it in solr with stored="false" and also store it in cassandra. If the format of how Solandra stores documents you don't want to have you change your application code. > Also, is index and query performance on par with a Sharded pure Solr > implementation? > > Since writes in Cassandra are durable the performance on writes isn't as fast as Solr's buffered writes on a single node. However since Solandra is multi-master the writes scale well across the cluster. Solandra perform on par with most solr searches one the data is in memory. Most people using Solandra are indexing TBs of data so they are IO bound... -Jake > Thanks for the feedback, > Ken > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/Anyone-familiar-with-Solandra-or-Lucendra-tp2927357p2953764.html > Sent from the Solr - User mailing list archive at Nabble.com. > -- http://twitter.com/tjake
